Enhanced Convolutional Neural Network for Accurate Crop Recommendation System on Climate Data

Abstract

A major factor in the nation's economic development and progress is agriculture. The primary cause of the significant decline in crop productivity is farmers' poor crop selection. However, the current approach struggles to estimate crop growth appropriateness only on a single characteristic, like soil or weather. Because of this, for the greatest and most reliable prediction, each of these aspects must be taken into account simultaneously. To enhance the overall system performance, this work proposes the Fusion of Lion Swarm Optimisation with Simulated Annealing (FLSOSA) and Enhanced Convolutional Neural Network (ECNN) algorithm. To boost crop productivity, a crop suggestion method based on the FLSOSA-ECNN algorithm is to be developed. This study's primary stages include crop forecast that is appropriate, FS (Feature selection), and pre-processing. The K-Nearest Neighbour (KNN) technique is employed for pre-processing, filling the values that are missing in the provided dataset. Then the pre-processed data is taken into attribute selection which is performed using FLSOSA algorithm. Utilizing an objective function, it is utilized to choose more relevant soil and meteorological parameters. Finally, these selected attributes are given into classification phase. In order to create a system that integrates the predictions of the FLSOSA-ECNN model and recommends the best crop based on the soil-specific type and attributes with a high degree of accuracy, the ECNN algorithm is utilised in this study for optimal crop prediction. According to the study findings, the suggested FLSOSA-ECNN methodology outperforms the current methods in terms of recall, accuracy, precision, and duration of execution.